timmy-config/scripts/pr_triage.py

#!/usr/bin/env python3
"""PR Triage Automation -- Categorize, deduplicate, report (#659)."""
import argparse, json, os, re, sys, subprocess
from collections import Counter, defaultdict
from datetime import datetime
from urllib.request import Request, urlopen
from urllib.error import HTTPError


def _token():
    t = os.environ.get("GITEA_TOKEN", "")
    if not t:
        p = os.path.expanduser("~/.config/gitea/token")
        if os.path.exists(p):
            t = open(p).read().strip()
    return t


def _api(url, token, method="GET", data=None):
    h = {"Authorization": "token " + token, "Accept": "application/json"}
    body = json.dumps(data).encode() if data else None
    if data:
        h["Content-Type"] = "application/json"
    req = Request(url, data=body, headers=h, method=method)
    try:
        return json.loads(urlopen(req, timeout=30).read())
    except HTTPError:
        return None


def fetch_prs(base, token, owner, repo):
    prs, page = [], 1
    while True:
        b = _api(base + "/api/v1/repos/" + owner + "/" + repo + "/pulls?state=open&limit=50&page=" + str(page), token)
        if not b:
            break
        prs.extend(b)
        if len(b) < 50:
            break
        page += 1
    return prs


def fetch_issues(base, token, owner, repo):
    iss, page = {}, 1
    while True:
        b = _api(base + "/api/v1/repos/" + owner + "/" + repo + "/issues?state=open&limit=50&page=" + str(page), token)
        if not b:
            break
        for i in b:
            if "pull_request" not in i:
                iss[i["number"]] = i
        if len(b) < 50:
            break
        page += 1
    return iss


def categorize(pr):
    c = (pr.get("title", "") + " " + pr.get("body", "") + " " + " ".join(l.get("name", "") for l in pr.get("labels", []))).lower()
    for kw, cat in [("training data", "training-data"), ("dpo", "training-data"), ("grpo", "training-data"),
                    ("fix:", "bug-fix"), ("bug", "bug-fix"), ("hotfix", "bug-fix"),
                    ("feat:", "feature"), ("feature", "feature"),
                    ("refactor", "maintenance"), ("cleanup", "maintenance"),
                    ("doc", "documentation"), ("test", "testing"), ("infra", "infrastructure")]:
        if kw in c:
            return cat
    return "other"


def refs(pr):
    return [int(m) for m in re.findall(r"#(\d+)", pr.get("title", "") + " " + pr.get("body", ""))]


def find_duplicates(prs):
    by = defaultdict(list)
    for p in prs:
        for r in refs(p):
            by[r].append(p)
    return [g for g in by.values() if len(g) > 1]


def health(pr, issues):
    r = refs(pr)
    created = datetime.fromisoformat(pr["created_at"].replace("Z", "+00:00"))
    updated = datetime.fromisoformat(pr["updated_at"].replace("Z", "+00:00"))
    now = datetime.now(created.tzinfo)
    return {
        "pr": pr["number"], "title": pr["title"], "head": pr["head"]["ref"],
        "category": categorize(pr), "refs": r,
        "open": [x for x in r if x in issues], "closed": [x for x in r if x not in issues],
        "age": (now - created).days, "stale": (now - updated).days,
        "mergeable": pr.get("mergeable"), "author": pr.get("user", {}).get("login", ""),
    }


def report(repo, checks, dups):
    lines = ["# PR Triage -- " + repo,
             "Generated: " + datetime.now().strftime("%Y-%m-%d %H:%M"),
             "Open PRs: " + str(len(checks)), "", "## Summary", ""]
    cats = Counter(h["category"] for h in checks)
    lines.append("| Category | Count |")
    lines.append("|----------|-------|")
    for c, n in cats.most_common():
        lines.append("| " + c + " | " + str(n) + " |")
    stale = [h for h in checks if h["stale"] > 7]
    lines.extend(["", "Stale (>7d): " + str(len(stale)),
                  "Duplicate groups: " + str(len(dups)), ""])
    if dups:
        lines.append("## Duplicates")
        for g in dups:
            rs = set()
            for p in g:
                rs.update(refs(p))
            lines.append("Issues " + ", ".join("#" + str(r) for r in sorted(rs)) + ":")
            for p in g:
                lines.append("  - #" + str(p["number"]) + ": " + p["title"])
        lines.append("")
    if stale:
        lines.append("## Stale (>7d)")
        for h in sorted(stale, key=lambda x: x["stale"], reverse=True):
            lines.append("- #" + str(h["pr"]) + ": " + h["title"] + " -- " + str(h["stale"]) + "d")
        lines.append("")
    lines.append("## All PRs")
    lines.append("| # | Title | Category | Age | Stale | Merge |")
    lines.append("|---|-------|----------|-----|-------|-------|")
    for h in sorted(checks, key=lambda x: x["pr"]):
        m = "Y" if h["mergeable"] else ("N" if h["mergeable"] is False else "?")
        s = str(h["stale"]) + "d" if h["stale"] > 7 else "-"
        lines.append("| " + str(h["pr"]) + " | " + h["title"][:50] + " | " + h["category"] +
                      " | " + str(h["age"]) + "d | " + s + " | " + m + " |")
    return chr(10).join(lines)


def main():
    p = argparse.ArgumentParser(description="PR Triage Automation")
    p.add_argument("--base-url", default="https://forge.alexanderwhitestone.com")
    p.add_argument("--owner", default="Timmy_Foundation")
    p.add_argument("--repo", default="")
    p.add_argument("--json", action="store_true", dest="js")
    p.add_argument("--output", default="")
    a = p.parse_args()
    token = _token()
    if not token:
        print("No token"); sys.exit(1)
    repo = a.repo
    if not repo:
        try:
            remote = subprocess.check_output(["git", "remote", "get-url", "origin"], text=True).strip()
            m = re.search(r"[/:](\w[\w-]*)/(\w[\w-]*?)(?:\.git)?$", remote)
            if m:
                a.owner, repo = m.group(1), m.group(2)
        except Exception:
            pass
    if not repo:
        print("No repo specified"); sys.exit(1)
    print("Triaging " + a.owner + "/" + repo + "...", file=sys.stderr)
    prs = fetch_prs(a.base_url, token, a.owner, repo)
    issues = fetch_issues(a.base_url, token, a.owner, repo)
    checks = [health(pr, issues) for pr in prs]
    dups = find_duplicates(prs)
    if a.js:
        print(json.dumps({"repo": repo, "prs": checks,
                          "duplicates": [[{"number": p["number"], "title": p["title"]} for p in g] for g in dups]},
                         indent=2))
    else:
        r = report(repo, checks, dups)
        print(r)
        if a.output:
            with open(a.output, "w") as f:
                f.write(r)
    print("\n" + str(len(checks)) + " PRs, " + str(len(dups)) + " duplicate groups", file=sys.stderr)


if __name__ == "__main__":
    main()